Hi YXI friends,

We start with a quick note on yesterday’s FOMC first.

Like the first, Kevin Warsh’s second FOMC Press Conference was a game of dodgeball. While we saw 3 dissents to the 9 votes to keep rates the same, Warsh offered little substance in his answers regarding the Fed’s internal debate or rationale.

The truth behind an afternoon of refusing to give answers is that Warsh wants the market to lead the path. Trillions of dollars are traded daily in the bond market - a high-stakes game watched keenly by the Fed. If the Fed is on the wrong course, the market will scream it. And that’s exactly the type of reaction Warsh wants to see. Markets are now beginning to directly digest real-time economic, geopolitical, and supply chain issues, rather than reacting to what they think the Fed might think. In short, Warsh doesn’t mind a smaller, quieter Fed that lets the market do the hard work.

I think that may be how Warsh passed his job interview. Bessent is of prominent hedge fund background and would have favoured the same approach. At the same time, by keeping his lips tight, if the Fed makes a rates decision that Trump doesn’t want to see, he could always point to the market reaction rather than his own words.

Turning to the Asset Price Cycles, and one thing worth making explicit before the readings. These fits update every week. The spectral analysis re-runs on a rolling five-year window, so as new data arrives the model can settle on a different dominant rhythm and the projected turn moves with it.

The most useful change is that the old cluster has dispersed. Two weeks ago most of the board pointed at the same trough about 68 trading days out; this week the projected turns spread from days to months, which makes the timing far more informative name by name. The bearish weight is still there, with the indices, semis and most of the Mag-7 in downcycles, but the pockets turning up are scattered across rates (TLT), power (OKLO, VST, CEG), software (INTU, CRM, IBM, ORCL) and consumer (DUOL, APP, TTD, UBER), while spot crypto has rolled over even as COIN and MSTR turn higher. Energy has gone the other way entirely, from the cleanest upcycle on the July board to troughing or falling in all three names.

DISCLAIMER: This newsletter is intended for educational purposes only. Any information or analysis in this note does not constitute an offer to sell or a solicitation of an offer to buy any securities. Nothing in this note is intended to be investment advice, nor should it be relied upon to make investment decisions. Any opinions, analyses, or probabilities expressed in this note are those of the author as of the note's date of publication and are subject to change without notice.

Please check out our glossary for explaining the acronyms, financial terminology, and economic data.

The macro surface flipped hard on Wednesday, July 29, 2026. Risk appetite fell from +0.18 on July 28 to -1.75, while inflation swung from -0.48 to +0.88 in a single session. Liquidity gave back the previous day's improvement, moving from +0.63 to -0.07.

I also happen to be a chart technician, and I can see that the risk scores (red line) are showing lower highs and lower lows since late June. Until liquidity turns positive again and risk recovers from -1.75, I favour not seeking new risk exposures.

Beneath every trend, markets tend to rise and fall in rough rhythms. They are the recurring swings that play out over weeks and months as positioning, liquidity and sentiment build up and unwind.

The charts below map those rhythms for each name I cover. Importantly, this is pure mathematics, not chart-reading or opinion. I run each price history through a spectral analysis (the same family of maths used to pick out the individual pitches hidden inside a sound wave).

It measures how strongly the price repeats at every possible cycle length and keeps the ones carrying the most energy. The long-run trend is stripped out first, so a name that has tripled isn't mistaken for one giant cycle.

On each chart:

A word on how to read them. Cycles are not clockwork, nor are they fixed laws of an asset. Their length and timing drift as the market regime changes, so treat each one as the dominant rhythm of the recent past rather than a guarantee.

I use them for context and timing, never as a standalone buy or sell signal. They explain only part of what price does, and a powerful trend or a fresh shock will override them. The real edge comes when a cycle lines up with the trend, the signals and the fundamentals.

How these are calibrated, for the technically minded:

IGV: Peaking (trough ~24d)
SOXX: Downcycle (trough ~15d)
SPY: Downcycle (trough ~35d)
IWM: Downcycle (trough ~39d)
QQQ: Downcycle (trough ~45d)

All five point down, but the timing has spread out. SOXX and IGV run the faster ~51-day wave and turn first, around 19 August and the start of September. SPY, IWM and QQQ share the ~171-day rhythm and bottom between mid-September and the start of October, several weeks earlier than the November window the July fit implied.

TLT: Upcycle (peak ~28d)

The clearest single change on the board. TLT read as a downcycle two weeks ago on the same ~171-day rhythm; the refreshed fit puts it in an upcycle with a projected peak in early September. On this read, rates now point the opposite way to equities rather than bottoming alongside them.

GLD: Peaking (trough ~104d)
SLV: Downcycle (trough ~31d)
USO: Downcycle (trough ~35d)

Gold has crossed from the rising side to the falling side of its ~171-day rhythm, though its projected trough sits far out in late December. Silver and oil both point to mid-September lows. The gold-silver split that stood out two weeks ago has closed: all three face the same direction now, just on very different schedules.

BTC: Downcycle (trough ~24d)
ETH: Downcycle (trough ~34d)

The pair has rolled over. Both read as troughing within eight sessions on the previous fit; both now sit in downcycles bottoming in early and mid-September, and they are no longer in lockstep. These remain two-cycle reads on a ~171-day rhythm, so treat the dates as indicative.

TSLA: Upcycle (peak ~43d)
META: Upcycle (peak ~13d)
GOOGL: Downcycle (trough ~10d)
NVDA: Downcycle (trough ~22d)
AMZN: Downcycle (trough ~26d)
AAPL: Downcycle (trough ~105d)
MSFT: Downcycle (trough ~139d)

The group has split much wider than the July fit showed. TSLA has crossed into upcycle with a peak projected for the end of September, and META is still rising but only about 13 sessions from its top. The five downcycles bottom anywhere from mid-August in GOOGL to mid-February in MSFT, so the shared November trough is gone.

QCOM: Peaking, still rising (peak ~160d)
AMD: Downcycle (trough ~12d)
MU: Downcycle (trough ~12d)
CRDO: Downcycle (trough ~33d)
TXN: Downcycle (trough ~35d)
AVGO: Downcycle (trough ~36d)
MRVL: Downcycle (trough ~39d)
TSM: Downcycle (trough ~43d)
ARM: Downcycle (trough ~52d)
INTC: Downcycle (trough ~55d)
ASML: Downcycle (trough ~60d)

Ten of eleven are in downcycles, the heaviest group on the board. The troughs ladder through from mid-August in AMD and MU out to late October in ASML rather than clustering, and CRDO and ARM are the two shortest histories in the set. QCOM is alone on the rising side, and its projected top sits a long way out, in March.

LRCX: Downcycle (trough ~12d)
AMAT: Downcycle (trough ~26d)
KLAC: Downcycle (trough ~61d)

All three point down, and each on a different rhythm. LRCX runs the fast ~57-day wave and bottoms around 14 August, AMAT the ~93-day wave into early September, and KLAC the long ~171-day wave which does not trough until late October. There is no shared date to work off here.

SNDK: Peaking, still rising (peak ~21d)
ANET: Troughing (trough ~16d)
WDC: Downcycle (trough ~13d)
CSCO: Downcycle (trough ~45d)
DELL: Downcycle (trough ~47d)

SNDK is the only name on the rising side, and it is a single-cycle fit on a short history, so the late-August peak carries less weight than the rest. WDC turns first on a ~51-day wave, around 17 August, with ANET close behind. CSCO and DELL share the long rhythm and do not bottom until the first week of October.

CRWV: Troughing (trough ~13d)
IREN: Troughing (trough ~171d)
EQIX: Downcycle (trough ~31d)
VRT: Downcycle (trough ~33d)
NBIS: Downcycle (trough ~39d)
HUT: Downcycle (trough ~52d)

The whole group points down. CRWV is nearest a turn at about 13 sessions, though that is a one-cycle fit on a short history. IREN's wave bottoms within days of here and then rises into the winter, so the ~171-day figure is counting the following trough rather than this one. The four downcycles are tighter than most of the board, bottoming between the second week of September and the middle of October.

OKLO: Upcycle (peak ~124d)
VST: Upcycle (peak ~113d)
TLN: Upcycle (peak ~23d)
CEG: Troughing, turning up (peak ~102d)
NRG: Troughing (trough ~13d)
CCJ: Troughing (trough ~18d)
GEV: Downcycle (trough ~57d)

One of the two most constructive groups on the board. OKLO and VST are early in long upcycles that do not top until January, and CEG has turned up with a peak projected for late December. TLN runs a much faster ~49-day wave and is already close to its high. NRG and CCJ are the exceptions, both falling into troughs inside the next month.

INTU: Upcycle (peak ~49d)
CRM: Upcycle (peak ~45d)
IBM: Upcycle (peak ~16d)
ORCL: Troughing, turning up (peak ~30d)
NOW: Peaking, still rising (peak ~6d)
SNOW: Peaking (trough ~97d)
ADBE: Downcycle (trough ~10d)
DDOG: Downcycle (trough ~44d)
PLTR: Downcycle (trough ~54d)

Five of nine sit on the rising side, which makes this the broadest turn-up on the board. INTU and CRM peak at the start of October, IBM in the third week of August, and ORCL has crossed up on a faster ~64-day wave. NOW is about six sessions from its top. The falling side is thinner by comparison, laddering from ADBE in mid-August out to PLTR in the middle of October.

CRWD: Peaking (trough ~70d)
PANW: Peaking (trough ~91d)
FTNT: Downcycle (trough ~13d)
NET: Downcycle (trough ~46d)
ZS: Downcycle (trough ~129d)

Nothing in the group is on the rising side. CRWD and PANW have only just rolled off their peaks, so their projected lows are still months out, in November and December. FTNT is the near-term turn, around 17 August, with NET at the start of October. ZS is the outlier, with a trough that does not arrive until February.

DUOL: Upcycle (peak ~27d)
APP: Upcycle (peak ~7d)
TTD: Troughing, turning up (peak ~72d)
UBER: Troughing, turning up (peak ~63d)
BKNG: Peaking, still rising (peak ~2d)
RDDT: Peaking (trough ~103d)
NFLX: Troughing (trough ~3d)

The most two-sided group here. TTD and UBER have both crossed the low, with peaks projected for late October and early November, and DUOL tops in early September. APP runs a much faster ~37-day wave and is already within a week of its high. At the other end BKNG is within days of its projected peak and NFLX within days of its trough, so both read as inflection rather than mid-move.

COIN: Upcycle (peak ~105d)
MSTR: Upcycle (peak ~104d)
HOOD: Peaking (trough ~41d)
MA: Troughing (trough ~16d)
V: Downcycle (trough ~60d)

COIN and MSTR sit in upcycles running to late December, which is the opposite of what spot BTC and ETH now show. The payment names face the other way, MA about 16 sessions from a low and V not until late October. HOOD is on a faster ~73-day rhythm and troughs in late September.

LNG: Troughing (trough ~5d)
CVX: Troughing (trough ~22d)
XOM: Downcycle (trough ~11d)

A full reversal from the July board, where these were the cleanest upcycles on the page. XOM has crossed into a downcycle, and LNG and CVX are both falling into troughs. The offset is timing: all three lows land inside the next month, LNG within a week.

FCX: Downcycle (trough ~22d)
NEM: Downcycle (trough ~55d)

Both down on the same ~171-day rhythm but a long way apart in timing, FCX in late August and NEM not until the middle of October. NEM sits on the same side as the GLD read above, so the metal and the miner point the same way.

ASTS: Peaking, still rising (peak ~19d)
RKLB: Downcycle (trough ~36d)

The two have separated. ASTS is still rising but only about 19 sessions from its projected top, while RKLB is already in a downcycle bottoming in mid-September.

Please help me improve the service with your immediate feedback - thank you.